<<< Date Index >>>     <<< Thread Index >>>

Re: maildir - is it just me or....



On Mon, Sep 01, 2003 at 07:39:25PM -0400, Ken Weingold wrote:
> I converted to maildir.  Is it just me or do maildir mailboxes take up
> a lot more disk space than mbox?  And the mailboxes seem to take
> longer to open, at least with over 1000 messages.

Yes.

Both problems are caused by the fact that each message is in one file.

The size is inflated substantially because with most filesystems
(ReiserFS being the only exception I can think of), each file uses
disk space in chunks equal to the block size.  So if your block size
is 4k, then each message will use at least 4k, even if it is only say
800 bytes.  That's a pretty substantial waste of disk space.  ReiserFS
will help out there, as it allows you to use fragments of blocks for
individual files.  But I've heard that ReiserFS is slow for some
common operations, so this may not help with the next issue.  On other
filesystems, your average disk waste will be about half a block per
message... not an insubstantial amount.

The speed is reduced because there is substantial overhead involved
with opening a file, and with a large number of messages in a maildir
mail folder, you have that many files to open.  It is much less
efficient to open multiple files, probably doing a disk seek in
between each one, than it is to simply open a file and read to the end
of it (as in the case of mbox), even taking into account disk
fragmentation.  As someone else commented, this problem is
exascerbated by the way many Unix-based filesystems handle many files
in a given directory.  But the problem is inherent to maildir,
regardless of the file system used -- it is only that much slower on
file systems which handle large numbers of files poorly.

As yet another person pointed out, maildir works quite well for
incoming mail spools, so long as you keep them relatively small.  For
archival purposes, mbox is generally better: you have not the problems
with locking, since mail isn't delivered there, and opening large mbox
mail folders is fast.  The former assumes, of course, that you are
moving messages into the archive folder manually, and no new messages
get delivered to them by the MDA.  If you are actually having mail
delivered there, particularly over NFS, you could still have a
problem.

-- 
Derek D. Martin
http://www.pizzashack.org/
GPG Key ID: 0xDFBEAD02
-=-=-=-=-
This message is posted from an invalid address.
Replying to it will result in undeliverable mail.
Sorry for the inconvenience.  Thank the spammers.

Attachment: pgpSJ2lJI2r6Y.pgp
Description: PGP signature